Re: Rack / Enclosure / Node Zero temperatures & cooling solutions

["Richard Boreham" <richard.boreham@xxxxxxxxxxxx>]

Martyn

=20

Just a few of thoughts on your set up and in no particular order=85

=20

-          One of the key factors in efficient cooling in the temperature
delta between the coolant (air) and the item(s) being cooled as this
create=
s
the temperature gradient that =91sucks=92 heat from the hot side.  So
getti=
ng
the cooling air as low as possible is a =91good thing=92 =96 hence chilled
=
systems
are intuitively better (although not necessarily overall more efficient /
cheaper due to the energy inefficiency of chilling process).  Maths is way
too difficult to contemplate (I can=92t even remember where to start now)
especially for a non separated system as you have when you blow air into a
cabinet ;-)

-          All that being said it looks like your garage floor is cooling
the ambient air temp and giving you a lower =91cool side=92 than you would
=
have
had in the house last year.  Hence there will be a lower overall exhaust
temp (assuming same heat load in the cabinet).=20

-          Have you tried switching off the lower blow in fan and free
venting from the base of the cabinet?  I ask for 2 reasons.  First, from
your temp readings (and I don=92t know exactly where the lower is placed)
y=
ou
may be creating more turbulence and back mixing in the cabinet between hot
and cool air at the top and bottom.  Just sucking from the top may reduce
this and improve the overall cooling effect.  Second, it would cut in half
your cooling (fan) costs and thus double the efficiency!  I stress here
=91may=92 because it=92s going to be very empirical and way too hard again
=
to be
definitive but might be worth a try!!!=20=20=20=20

-          My approach is a bit different=85 I have a node 0 in loft that
suffers massive solar gain (black roof).  I noticed the loft temp reached
4=
5
deg C recently (and I=92m going to watch with interest today / tomorrow). 
=
I
suck air from the soffit (around 20 - 22 deg) and blow this directly into
the PC and UPS inlet vents.  Temp in the rest of the cabinet has then maxed
at about 40deg C but this is a combination temp of the loft air and the air
vented from the PC/UPS and a big 12 channel power amp.  Unfortunately I
don=92t have a temp sensor in the PC & UPS enclosures but I=92d
estimate th=
at
the temp in them is prob around 30-35 deg max =96 components like drives
an=
d
processors will obviously be higher.  Overall a bit higher than I=92d
reall=
y
like but it seems to be ok in practice.  If I did it all again, I=92d not
p=
ut
a node 0 in the loft without it being insulated under the rafters. That was
our original plan but a cost reduction exercise hit that bit of our build =
=96
I reckon it=92s about a =A32k - =A32.5k job in materials alone to do that
j=
ob
properly and is still on the to-do list!=20=20=20

=20

Rgds

=20

Richard



[Non-text portions of this message have been removed]